Whitepaper
Keystone Finance · Whitepaper v1.1 · 2026 · Kamran Choudhry
ksUSD earns the carry Solana's markets already produce: perp funding, jitoSOL staking, and USDC lending. Deposit USDC, hold ksUSD, and the share price rises as that carry accrues. No token emissions, no off-chain Treasury bills, and every position is on-chain.
One program, one vault, one share mint. Over the recent 24-month backtest the vault returns ~7% net APY (~9% gross) with a −1.2% max drawdown, flooring near ~4% when funding is thin.
I. What ksUSD is
Solana's markets pay carry continuously: perps settle funding hourly, jitoSOL compounds each epoch, lending reprices each slot. The dollars sitting on top of it collect none of that, and the few that pay yield source it off-chain. ksUSD collects the on-chain carry directly, hedged so SOL price moves cancel, and passes it through the share price.
II. How it works
ksUSD runs a delta-neutral basis trade with two modes, set by the funding rate. Transitions are automatic: the on-chain smoothed funding signal decides the mode and a keeper bot executes the switch, filtered by a 7-day funding mean and a 12-hour minimum hold. Net SOL exposure stays near zero in normal basis.
Normal basis
funding ≥ +2%
long jitoSOL spot (unlevered) + short SOL-PERP on Phoenix at 1×, USDC margin
jitoSOL staking + funding received
Parked
otherwise
all capital in USDC lending (Kamino)
lending yield (~4–5%)
The +2% threshold is a risk buffer, not a break-even line: on pure yield the basis beats parked down to about −1.3% funding, because jitoSOL staking (~5.8%) already tops USDC lending (~4%). Setting entry at +2% leaves a ~3-point margin above that break-even — the premium the vault demands before taking on perp-venue, liquidation, and switch-churn risk. Below it, ksUSD parks in USDC lending rather than run the basis for a thin, risk-unadjusted spread.
Reverse basis (harvesting negative funding by borrowing and shorting the underlying) is built but dormant in v1. Negative funding and the borrow rate tend to spike together, so the net rarely clears cost; Parked covers that range instead. It turns on only if the funding-minus-borrow spread clears a set hurdle. See Strategy & Modes.
III. The ksUSD share
ksUSD is a non-rebasing share token. Your balance is fixed, and the share price rises as carry accrues. You redeem against NAV: instant from the liquidity buffer, queued for larger size.
It is a share on a hedged carry vault, not a $1-pegged stablecoin. It carries no fixed yield, holds no RWA or T-bills, pays no emissions, and takes no view on SOL price.
The architecture is small on purpose: one program, one vault, one rule set.
Yield sources & fees
Phoenix funding
Normal basis
5–15% APR
jitoSOL staking
Normal basis (spot leg)
~5.8% APR
USDC lending
Buffer + reserve + parked
~4–5% APR
Management
0%
Performance
20% above HWM (net-new gains only)
Reserve skim
5% of perf → on-chain reserve fund
Withdrawal
0%
IV. Why Solana
The design needs a few things on one chain, close enough to compose:
An on-chain perp. Phoenix Perps settles on-chain, so the vault opens its hedge by calling into it with USDC margin from its own program. That direct composition is the Solana-specific part.
Funding it receives, rather than a pool borrow-fee it pays.
A high-yield LST and USDC lending: jitoSOL held unlevered for staking, Kamino for the buffer and parked NAV.
Fees small enough that regular rebalancing stays economic.
On Ethereum these don't line up: gas makes rebalancing expensive, the deepest perps are off-chain or on separate domains, and LST yield, funding, and lending live apart. The spot leg also earns more here. Solana staking plus MEV tips put jitoSOL near ~5.8%, above Ethereum LSTs' ~3%, so the unlevered leg carries more before any composability or fee edge.
V. Performance — daily backtest (Jul 2024 – Jun 2026, 24 months)
Phoenix launched Dec 2025, so the backtest runs on historical SOL-perp funding as a proxy. Over the recent 24 months — which span the 2025–26 funding compression — the v1 set (normal + parked) compounded +14.5% net ($100 → $114.50), parking 22% of days when funding didn't pay. Reverse is excluded.
Normal + parked (v1)
~7%
~9%
$114.50
−1.2%
sUSDe benchmark
~5% recent
—
—
bleeds in low funding
USDC lending benchmark
~4%
—
—
flat floor
The ~7% net is fully loaded: it already carries the 20% performance fee, a margin-capital haircut (USDC posted as perp margin at ~10× leverage earns no basis), and trading costs. USDC lending (~4%) is the benchmark — the same capital's alternative — and the vault's parked floor sits about there. The real result is the −1.2% max drawdown through a funding collapse, not the yield magnitude.
Methodology: funding is an on-chain SOL-perp archive for 188 of 730 days (the archive ends Jan 2025, when the venue went down) plus daily Binance funding × empirical ratio for the rest. Phoenix anchors its own funding to CEX price feeds (Binance, Coinbase, Hyperliquid, Bybit), so the Binance proxy tracks the same prices Phoenix references, not an unrelated market. 7-day mean for mode classification; margin at ~10× leverage, below Phoenix's verified 15× venue max, so over-margined (9.1% of NAV at 0% yield); 5 bps perp fee + 10 bps slippage per side + 20–40 bps per switch. Under stress a month can run worse than the modeled −1.2%. Detail: historical-simulation.md.
VI. Resilience — redemption and risk
Share price is NAV ÷ shares, marked from on-chain balances and oracle-priced legs. Keeper attestation is bounded by a per-hour change cap and an oracle sanity band, so it can't mark NAV past what the chain shows. A 10%-of-NAV buffer covers normal redemptions instantly; larger ones burn at the locked price and settle against realized unwind proceeds through a permissionless crank. In stressed markets a queued payout can come in below marked NAV.
Smart-contract
Pre-mainnet audit scheduled; devnet only until then.
Perp-venue dependency
Phoenix is the only hedge venue, so an exploit, outage, or socialized loss hits ksUSD directly. Phoenix is in private beta, and trader onboarding needs an Ellipsis builder-access grant, a live external dependency.
Counterparty
Phoenix, Ember, Kamino, Jupiter, Jito. No single one holds all NAV; the reserve fund absorbs first loss up to its size.
Oracle
Pyth reads gated by 5-min staleness and 2% confidence; outages past that can still cause loss.
Funding compression
Parked earns only ~4%, so the ~7% isn't guaranteed in a flat regime.
jitoSOL depeg
Marked at the stake-pool rate; a 5% depeg auto-pauses and arms emergency_close. Slashing is unhedgeable.
Liquidation
The Phoenix short is over-margined so a SOL rally is absorbed before rebalancing.
Keeper outage
Buffer withdrawals stay open; rotation and queue processing pause until cranking resumes. Anyone can crank.
Fallbacks are built in. Funding compression parks the vault. Perp-venue trouble triggers an oracle-divergence auto-pause and a permissionless emergency_close. A depeg auto-pauses. Wind-down turns the vault into a pro-rata USDC claim. Redemption works whether or not the team is present.
VII. Position and outlook
ksUSD
Perp funding + staking + lending
Positive-funding basis + parked floor
Fully on-chain
Phoenix, Kamino, Jito (Solana-native)
Ethena (sUSDe)
Delta-hedged perp funding
Long-basis only
Off-chain CEX, attested
CEXs + custodians
MakerDAO / Sky
Stability fees + RWA/savings
n/a
On-chain + off-chain RWA
RWA counterparties
Perena (USD*)
Swap fees + stable yields
n/a
On-chain
Solana AMM/DEX
Ethena showed the demand for a yield-bearing dollar, and the limit of a one-directional one: through 2025 sUSDe rode positive funding to ~15%, then fell to ~3.7% as funding compressed.¹ ksUSD earns the basis when funding pays and parks when it doesn't.
The longer goal is a dollar whose yield comes from, and can be checked against, the market it lives on.
Simulated performance does not predict future results.
¹ Ethena figures are directional, per public dashboards from the backtest era.
app.keystonefi.xyz · docs.keystonefi.xyz · Simulated figures only. Not investment advice.
Last updated